osint-dashboard/tests/test_api_news.py

"""Integration tests for the news pipeline API (GET /api/news + summaries + intel).

DB-backed: marked `requires_db` and auto-skip when the test database is
unreachable (see tests/conftest.py). Seeding writes directly to the shared
`articles` / `article_summaries` / `news_items` tables, exactly as the scraper
+ summarizer services would.
"""

from __future__ import annotations

import asyncio
import os
from datetime import datetime, timezone

import asyncpg
import httpx
import pytest

from conftest import requires_db

from main import app

BASE = "http://test"


def _conn_kwargs() -> dict:
    return {
        "host": os.environ["DB_HOST"],
        "port": int(os.environ["DB_PORT"]),
        "user": os.environ["DB_USER"],
        "password": os.environ["DB_PASSWORD"],
        "database": os.environ["DB_NAME"],
    }


def _truncate() -> None:
    async def run():
        conn = await asyncpg.connect(**_conn_kwargs())
        try:
            await conn.execute(
                "TRUNCATE articles, article_summaries, news_items CASCADE"
            )
        finally:
            await conn.close()

    asyncio.run(run())


@pytest.fixture()
def clean_news():
    _truncate()
    yield
    _truncate()


def _seed_article(title: str, url: str, domain: str, ts: str, content: str = "body text") -> None:
    async def run():
        conn = await asyncpg.connect(**_conn_kwargs())
        try:
            await conn.execute(
                "INSERT INTO articles (title, url, content, domain, timestamp) "
                "VALUES ($1, $2, $3, $4, $5)",
                title, url, content, domain, datetime.fromisoformat(ts),
            )
        finally:
            await conn.close()

    asyncio.run(run())


def _seed_summary(text: str, ts: str, model: str | None = None, kind: str | None = None) -> int:
    async def run() -> int:
        conn = await asyncpg.connect(**_conn_kwargs())
        try:
            row = await conn.fetchrow(
                "INSERT INTO article_summaries (summary_text, batch_timestamp, model, kind) "
                "VALUES ($1, $2, $3, $4) RETURNING id",
                text, datetime.fromisoformat(ts), model, kind,
            )
            return int(row["id"])
        finally:
            await conn.close()

    return asyncio.run(run())


def _seed_news_item(
    summary_id: int,
    kind: str,
    headline: str,
    importance: str,
    *,
    location_name: str | None = None,
    lat: float | None = None,
    lon: float | None = None,
    location_confidence: str | None = None,
    category: str | None = None,
    url: str | None = None,
) -> None:
    async def run():
        conn = await asyncpg.connect(**_conn_kwargs())
        try:
            await conn.execute(
                "INSERT INTO news_items "
                "(summary_id, kind, headline, importance, location_name, "
                " lat, lon, location_confidence, category, url) "
                "VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10)",
                summary_id, kind, headline, importance, location_name,
                lat, lon, location_confidence, category, url,
            )
        finally:
            await conn.close()

    asyncio.run(run())


def _get(path: str) -> httpx.Response:
    async def _get_async() -> httpx.Response:
        transport = httpx.ASGITransport(app=app)
        async with httpx.AsyncClient(transport=transport, base_url=BASE) as client:
            return await client.get(path)

    return asyncio.run(_get_async())


@requires_db
def test_api_news_returns_newest_first(clean_news):
    _seed_article("older", "https://a.example/1", "a.example", "2026-08-24T17:00:00+00:00")
    _seed_article("newer", "https://a.example/2", "a.example", "2026-08-24T18:00:00+00:00")
    body = _get("/api/news").json()
    assert isinstance(body, list)
    assert len(body) == 2
    # newest first
    assert [a["title"] for a in body] == ["newer", "older"]
    # exact JSON contract the frontend news panel needs
    assert set(body[0].keys()) == {
        "id", "title", "url", "content", "domain", "timestamp",
    }
    assert body[0]["domain"] == "a.example"


@requires_db
def test_api_news_domain_filter(clean_news):
    _seed_article("x", "https://x.example/1", "x.example", "2026-08-24T17:00:00+00:00")
    _seed_article("y", "https://y.example/1", "y.example", "2026-08-24T17:00:00+00:00")
    body = _get("/api/news?domain=y.example").json()
    assert len(body) == 1
    assert body[0]["domain"] == "y.example"


@requires_db
def test_api_news_since_filter(clean_news):
    _seed_article("before", "https://a.example/1", "a.example", "2026-08-24T17:00:00+00:00")
    _seed_article("after", "https://a.example/2", "a.example", "2026-08-24T18:30:00+00:00")
    body = _get("/api/news?since=2026-08-24T18:00:00Z").json()
    assert len(body) == 1
    assert body[0]["title"] == "after"


@requires_db
def test_api_news_summaries_contract(clean_news):
    _seed_summary("master summary markdown…", "2026-08-24T18:05:00+00:00")
    body = _get("/api/news/summaries").json()
    assert isinstance(body, list)
    assert len(body) == 1
    s = body[0]
    assert set(s.keys()) == {"id", "summary_text", "batch_timestamp", "model", "kind"}
    assert s["summary_text"] == "master summary markdown…"
    assert s["batch_timestamp"].startswith("2026-08-24T18:05")
    assert s["kind"] is None


@requires_db
def test_api_news_summaries_kind_filter(clean_news):
    _seed_summary("interval brief", "2026-08-28T22:05:00+00:00", kind="interval")
    _seed_summary("daily recap", "2026-08-28T03:00:00+00:00", kind="daily_recap")
    recap = _get("/api/news/summaries?kind=daily_recap").json()
    assert len(recap) == 1
    assert recap[0]["summary_text"] == "daily recap"
    assert recap[0]["kind"] == "daily_recap"
    assert _get("/api/news/summaries?kind=nope").status_code == 422


@requires_db
def test_api_news_empty(clean_news):
    assert _get("/api/news").json() == []
    assert _get("/api/news/summaries").json() == []
    assert _get("/api/news/ticker").json() == []
    assert _get("/api/news/map").json() == []


TICKER_KEYS = {"id", "headline", "importance", "location_name", "url", "created_at"}
MAP_KEYS = {
    "id", "headline", "importance", "location_name", "lat", "lon",
    "location_confidence", "category", "url", "created_at",
}


def _seed_flagged_items() -> None:
    sid = _seed_summary("batch brief", "2026-08-27T18:05:00+00:00", "Hermes-4.3-36B")
    _seed_news_item(
        sid, "ticker", "Critical ticker", "critical",
        location_name="Kyiv", url="https://example.com/ticker",
    )
    _seed_news_item(
        sid, "ticker", "Low ticker", "low",
        location_name="Somewhere", url="https://example.com/low",
    )
    _seed_news_item(
        sid, "map", "Critical map", "critical",
        location_name="Taipei", lat=25.03, lon=121.56,
        location_confidence="high", category="conflict",
        url="https://example.com/map",
    )


@requires_db
def test_api_news_ticker_returns_only_flagged(clean_news):
    _seed_flagged_items()
    resp = _get("/api/news/ticker")
    assert resp.status_code == 200
    body = resp.json()
    assert isinstance(body, list)
    assert len(body) == 1
    item = body[0]
    assert set(item.keys()) == TICKER_KEYS
    assert item["headline"] == "Critical ticker"
    assert item["importance"] == "critical"
    assert item["location_name"] == "Kyiv"
    assert item["url"] == "https://example.com/ticker"


@requires_db
def test_api_news_ticker_falls_back_to_lesser_when_nothing_flagged(clean_news):
    sid = _seed_summary("quiet brief", "2026-08-27T18:05:00+00:00", "Hermes-4.3-36B")
    _seed_news_item(
        sid, "ticker", "Shop theft downtown", "low",
        location_name="Raleigh", url="https://example.com/theft",
    )
    resp = _get("/api/news/ticker")
    assert resp.status_code == 200
    body = resp.json()
    assert len(body) == 1
    assert body[0]["headline"] == "Shop theft downtown"
    assert body[0]["importance"] == "low"


@requires_db
def test_api_news_map_returns_only_flagged_with_coords(clean_news):
    _seed_flagged_items()
    resp = _get("/api/news/map")
    assert resp.status_code == 200
    body = resp.json()
    assert isinstance(body, list)
    assert len(body) == 1
    item = body[0]
    assert set(item.keys()) == MAP_KEYS
    assert item["headline"] == "Critical map"
    assert item["importance"] == "critical"
    assert item["lat"] == 25.03
    assert item["lon"] == 121.56
    assert item["location_confidence"] == "high"
    assert item["category"] == "conflict"
    assert _get("/api/news/map?bbox=1,2,3").status_code == 422